Work summary for the end of and work plan for the end of, end

Work summary for the end of and work plan for the end of, end

Having become accustomed to summing up and planning, we can't systematize our scattered experiences without summing up. In the end, we can only see that trees do not see the forest. Without a plan, we can have no goals and will be decadent if there is no target, in the end, Wang maofa posted Spring Festival couplets for one year and one year. This is especially true for O & M. O & M is self-driven, while development is demand-driven, which is quite different: O & M involves a wide range of knowledge, and the specific work is also an extremely scattered and unpredictable emergency, which even makes you unprepared. Without summing up the work, it is very painful and always plays the role of a fireman; after several years of development, most of the work of the O & M infrastructure system has been automated through scripts and systems, freeing up the repetitive work of O & M to a great extent. At this time, it is easy to do nothing and feel at ease, this is profound understanding. All plans are even more important for O & M. They plan to improve themselves and plan the optimization system... They all determine their future!

The following is a summary of my work emails (only replacing sensitive areas). I encourage myself and welcome to make a picture.

Over the years, we will summarize our work in 2014 and plan our work objectives in 2015.

I. O & M management is mainly completed:

(1) The platform launched the mysql slow query analysis push system by day, and automatically sent the results after category analysis to relevant personnel;

(2) hadoop knowledge learning and research, after taking over hadoop, enriched the monitoring and warning project, added the automatic capacity management and notification functions, and made a small number of Optimizations to the cluster;

(3) creatively implements layer-7 (http & https) NAT forwarding for multiple Intranet sites with a single public IP address based on a domain name, after the Intranet web application test is completed, the history of IP + port needs to be remembered (the application is in the dob * and part of the test environment of Zhang );

(4) The overall relocation of Shi ** data center to ** data center and Site Selection of the data center in the early stage;

(5) The online monitoring system has added support for multiple redis instances and multiple proxy servers;

(6) ** learn and monitor the freight call center system;

(7) Expansion of the offline vm system and separation of ** cargo-related VM resources;

(8) offline Release System rewriting. Combined with svn, the system has perfectly implemented a fully automated code release notification system.

II. Specific business aspects have been completed:

(1) upgrading and resizing of important servers (adding dual-power supplies, resizing from the database SSD disk, and adding backup devices for servers/switches );

(2) Archiving and cleaning of historical online business data (tra ** database and mongodb database );

(3) platform image server migration;

(4) redis, nginx proxy, and haproxy Single Point solution (for the time being, it is cold switching. The redis high availability solution has been completed and runs stably for more than two months online, and is launched as early as 2015 );

(5) sphinx performance optimization: improve the performance of the Word Segmentation service by adding multiple instances and dedicated slave databases, excluding the effect of freezing and blocking caused by sphinx index refreshing on other services );

(6) online and offline deployment and automatic release of new services (MP service, Upload service, impact *, dob **, etc );

(7) other daily O & M work (this is a bit complex and will not be listed ).

Iii. Emergency Handling:

(1) ** when problems such as the IDC and company's network, server load, application, and DB affect normal access, the O & M team should promptly troubleshoot, fix, follow up, and notify relevant business parties;

(2) timely handling of some server hardware problems (such as bad track of the master database disk, hadoop disk failure, and failure of the datanode motherboard to start up );

(3) Check, confirm, and upgrade all related systems as soon as the heartbleed and bash vulnerabilities are detected in openssl. (For details, see "Notice on bash upgrade for all linux servers of the company 20140925" in the email.)

Iv. Personal aspects:

(1) Red Hat RHCA certification has been obtained through learning.

V. Main Work Plan for 2015 (if continued ):

(1) Establish a centralized management system for offline accounts (unified management of personal server accounts, jira, svn, and springboards). There is no need to redo online bastion hosts to manage accounts;

(2) establish a bastion host system (self-developed) with a unified entrance for offline servers );

(3) Pay attention to the running status of old servers, and propose the implementation scheme of hardware upgrade in a timely manner;

(4) solution to single point of service (For details, refer to "Platform service single point of failure sorting 20141114");

(5) As the system becomes increasingly complex, offline devbox cannot meet the needs of developers. Xiong * and li * discuss this issue in the early stage. If it is feasible, we will work with Xiong.

EOF